Azure Cosmos DB
Connection and object support require review
Record types to review with Stacksync
| Record types | Coverage and requirements |
|---|---|
| Confirm support for this record type and the direction you need. |
Database / AI · Two-way sync platform
Azure Cosmos DB and Azure OpenAI integration
Plan how Azure Cosmos DB and Azure OpenAI should share data across your business. Work with Stacksync engineers on record mapping, system access, and the requirements for running the integration.
- Scope your workflow with an integration engineer
- Review the systems, records, and updates you need
Built for teams where self-serve, reliability and scale matter
Case study
Migrated from MuleSoft
Case study
Migrated from Celigo
Migrated from Heroku Connect
Migrated from Matillion
Case study
Migrated from Fivetran
Case study
Migrated from Celigo
Proposed workflow
File or document metadata sync test
Planning example. Stacksync support for the required connection and record operations needs a technical review.
Starting eventA change involving Azure Cosmos DB Items (JSON documents) or Azure OpenAI Files needs a defined result in the other system.
Start with Azure Cosmos DB Items (JSON documents) and Azure OpenAI Files. Use the record-matching and field-ownership rules from your mapping worksheet.
Resolve folder/container and parent-record references before attaching metadata.
Test a normal update and one failed or repeated update in the supported direction. Keep both record IDs with the test results.
What to verifyTest a renamed file, a new version, a moved folder, and an access-restricted document.
Review records and field ownershipProposed record relationships
Records to connect
Use these examples to define record matching and field ownership for your technical review.
Download the mapping worksheetCSV · No email required
| Azure Cosmos DB record | Azure OpenAI record | Record matching | Field ownership |
|---|---|---|---|
| Items (JSON documents)Proposed record; confirm Stacksync object support.Record matching | FilesProposed record; confirm Stacksync object support. | Keep file/object ID, container, and version. A path can change and a filename can repeat. | Separate document content, metadata, and sharing permissions; a metadata sync does not imply file transfer or ACL replication. |
| Proposed file or document metadata tableProposed table; choose its name and schema.Reporting dataset | FilesProposed record; confirm Stacksync object support. | Keep file/object ID, container, and version. A path can change and a filename can repeat. | Separate document content, metadata, and sharing permissions; a metadata sync does not imply file transfer or ACL replication. |
| Proposed model or deployment metadata tableProposed table; choose its name and schema.Reporting dataset | ModelsProposed record; confirm Stacksync object support. | Keep provider/project, model or deployment ID, and version distinct. | The AI platform owner controls model deployment; metadata movement does not invoke a model or authorize model changes. |
These relationships do not establish connector availability. Review the required connection and record operations with Stacksync.
Record coverage to review
Use documented coverage where available. Catalog record types are starting points for review and do not confirm Stacksync support.
Azure OpenAI
Connection and object support require review
Record types to review with Stacksync
| Record types | Coverage and requirements |
|---|---|
| Confirm support for this record type and the direction you need. |
Connection essentials
Confirm Stacksync support and account requirements for undocumented connections. Interface information alone does not establish connector availability.
View setup requirements and limits| Connection requirement | Azure Cosmos DB | Azure OpenAI |
|---|---|---|
| Integration interface | REST API and SDKs over HTTPS (API for NoSQL, formerly the SQL API); also MongoDB, Cassandra, Gremlin, and Table API surfaces | REST data-plane (inference + authoring) and Azure Resource Manager control-plane |
| Authentication | Confirm the credentials, API plan, and permissions required for Azure Cosmos DB. | Confirm the credentials, API plan, and permissions required for Azure OpenAI. |
| Change detection | Confirm how Stacksync detects changes for this connector and the objects you need. | Confirm how Stacksync detects changes for this connector and the objects you need. |
| Read access | Confirm with Stacksync | Confirm with Stacksync |
| Write access | Confirm with Stacksync | Confirm with Stacksync |
Enterprise controls
Security and control for your integrations
Explore security controlsCompliance and data transfers
As a data company, we understand the importance of keeping your data secure. Stacksync is built with security best practices to keep your data safe at every layer, and is DPF-certified for US, EU, UK and CH data transfers.
- SOC 2 Type II
- ISO 27001
- HIPAA BAA
- GDPR
- CCPA
- DPF US-EU-UK-CH
SSO & SCIM
Let your users access Stacksync from your centralized user management systems. Works with Okta, Azure, Google SSO and more.
Alerts
Immediately get alerted about record syncing issues over email, Slack, PagerDuty and WhatsApp. Resolve issues from a centralized dashboard with retry and revert options.
Secure connection options
Securely connects to your systems with:
Record-level recovery
Inspect sync errors and use retry and revert controls to resolve failed updates.
Read the recovery guideImplementation
Technical reference
Review setup, record relationships, testing, and recovery for your implementation.
Authentication, permissions and API limits
Connection requirements and limits
Azure Cosmos DB
- Integration interface
- REST API and SDKs over HTTPS (API for NoSQL, formerly the SQL API); also MongoDB, Cassandra, Gremlin, and Table API surfaces
- Authentication
- Confirm the credentials, API plan, and permissions required for Azure Cosmos DB.
- Change detection
- Confirm how Stacksync detects changes for this connector and the objects you need.
- Read access
- Confirm with Stacksync
- Write access
- Confirm with Stacksync
Setup requirements
- Identify the Azure Cosmos DB account, edition, environment, and business objects the integration must access.
- Confirm a Stacksync connector or implementation path for Azure Cosmos DB, including read/write support, authentication, and initial-load limits.
Limitations to check
- Confirm Stacksync support for Azure Cosmos DB and the record types your workflow needs.
- Review write-back, deletion handling, update timing, and account limits with the integration team.
Azure OpenAI
- Integration interface
- REST data-plane (inference + authoring) and Azure Resource Manager control-plane
- Authentication
- Confirm the credentials, API plan, and permissions required for Azure OpenAI.
- Change detection
- Confirm how Stacksync detects changes for this connector and the objects you need.
- Read access
- Confirm with Stacksync
- Write access
- Confirm with Stacksync
Setup requirements
- Identify the Azure OpenAI account, edition, environment, and business objects the integration must access.
- Confirm a Stacksync connector or implementation path for Azure OpenAI, including read/write support, authentication, and initial-load limits.
Limitations to check
- Confirm Stacksync support for Azure OpenAI and the record types your workflow needs.
- Review write-back, deletion handling, update timing, and account limits with the integration team.
Prepare Azure Cosmos DB and Azure OpenAI access
Set up both accounts before testing the mapping. Use test records where available, and identify the account administrator who can approve access and help resolve setup errors.
Azure Cosmos DB setup checklist
- Identify the Azure Cosmos DB account, edition, environment, and business objects the integration must access.
- Confirm a Stacksync connector or implementation path for Azure Cosmos DB, including read/write support, authentication, and initial-load limits.
Azure OpenAI setup checklist
- Identify the Azure OpenAI account, edition, environment, and business objects the integration must access.
- Confirm a Stacksync connector or implementation path for Azure OpenAI, including read/write support, authentication, and initial-load limits.
Prepare to go live
Record the fields each system can update, the first-load cutoff, both record IDs, the expected update delay, and who handles errors. Complete the tests before production before expanding to more records.
Use the Azure Cosmos DB and Azure OpenAI planning worksheet to capture these decisions. Record the access owner in the worksheet and enter credentials only in the connection setup.
Record identity and field ownership
Use these data-model references to describe the records your connection needs. They are planning examples; connector availability and supported operations must be established before implementation.
Download the mapping worksheet · CSV, no email required
Record matching
Items (JSON documents) / Files
Compare whether Azure Cosmos DB Items (JSON documents) and Azure OpenAI Files describe the same file or document metadata in your business.
Planning example. Stacksync support for the required connection and record operations needs a technical review.
- Azure Cosmos DB
- Object support to establish
- Azure OpenAI
- Object support to establish
- Record identity
- Keep file/object ID, container, and version. A path can change and a filename can repeat.
- Field ownership
- Separate document content, metadata, and sharing permissions; a metadata sync does not imply file transfer or ACL replication.
Fields to include
- Source file ID
- Container reference
- Version
- Metadata
- Access classification
- Record dependencies
- Resolve folder/container and parent-record references before attaching metadata.
- Validation
- Test a renamed file, a new version, a moved folder, and an access-restricted document.
- Recovery
- Check the current file version and destination access before retrying transfer or metadata updates.
Reporting dataset
Proposed file or document metadata table in Azure Cosmos DB (choose its name) / Files
Plan a file or document metadata dataset while preserving its source meaning.
Planning example. Stacksync support for the required connection and record operations needs a technical review.
- Azure Cosmos DB
- Your database schema
- Azure OpenAI
- Object support to establish
- Record identity
- Keep file/object ID, container, and version. A path can change and a filename can repeat.
- Field ownership
- Separate document content, metadata, and sharing permissions; a metadata sync does not imply file transfer or ACL replication.
Fields to include
- Source file ID
- Container reference
- Version
- Metadata
- Access classification
- Record dependencies
- Resolve folder/container and parent-record references before attaching metadata.
- Validation
- Test a renamed file, a new version, a moved folder, and an access-restricted document.
- Recovery
- Check the current file version and destination access before retrying transfer or metadata updates.
Reporting dataset
Proposed model or deployment metadata table in Azure Cosmos DB (choose its name) / Models
Plan a model or deployment metadata dataset while preserving its source meaning.
Planning example. Stacksync support for the required connection and record operations needs a technical review.
- Azure Cosmos DB
- Your database schema
- Azure OpenAI
- Object support to establish
- Record identity
- Keep provider/project, model or deployment ID, and version distinct.
- Field ownership
- The AI platform owner controls model deployment; metadata movement does not invoke a model or authorize model changes.
Fields to include
- Model/deployment reference
- Provider/project
- Version
- Lifecycle state
- Record dependencies
- Resolve project, access policy, model version, and any evaluation requirements.
- Validation
- Test a retired model, a renamed deployment, and an unauthorized project reference.
- Recovery
- Confirm the current deployment version and access before retrying a dependent action.
Compare integration approaches
Choose a method around one example record and the update your business needs. Use Items (JSON documents) / Files to review record matching and confirm Stacksync support for the required operations. Compare ongoing sync, a custom workflow, and a scheduled export against that requirement.
Stacksync managed sync
- Best fit
- Review compatibility with a Stacksync engineer using an example of the records and updates you need.
- Operating responsibility
- Fits ongoing record synchronization when the required operations are supported. Add workflow steps for approvals or business actions that go beyond copying fields.
- Before you choose
- Check record matching: Keep file/object ID, container, and version. A path can change and a filename can repeat. Verify field coverage, deletion handling, and how changes are detected.
Native vendor integration
- Best fit
- A vendor-built integration may fit if it supports your Azure Cosmos DB and Azure OpenAI record types.
- Operating responsibility
- Can reduce setup for a supported workflow. You may need another method for records or business steps it does not cover.
- Before you choose
- First check whether either vendor offers this integration. If available, verify Items (JSON documents) / Files, update direction, account tier, and related-record handling.
Custom API or workflow
- Best fit
- Consider when Azure Cosmos DB and Azure OpenAI need a transformation, approval, or action outside a direct record sync.
- Operating responsibility
- Provides control over business steps; the team owns credentials, version changes, error queues, and reconciliation.
- Before you choose
- Verify endpoint permissions, pagination, quotas, duplicate detection, and failure recovery.
File or scheduled snapshot
- Best fit
- Consider for a one-time Azure Cosmos DB / Azure OpenAI migration or a reporting need with an explicit freshness window.
- Operating responsibility
- Can simplify a bounded transfer; later changes and deletion history require another extraction or a separately designed incremental process.
- Before you choose
- Record the extraction cutoff, source IDs, encoding, date/number formats, and reconciliation totals.
Workflow scenarios and expected results
File or document metadata sync test
Planning example. Stacksync support for the required connection and record operations needs a technical review.
Starting event: A change to the selected Items (JSON documents) or Files record needs a defined result in the other system.
- Start with Azure Cosmos DB Items (JSON documents) and Azure OpenAI Files. Use the record-matching and field-ownership rules from your mapping worksheet.
- Resolve folder/container and parent-record references before attaching metadata.
- Test a normal update and one failed or repeated update in the supported direction. Keep both record IDs with the test results.
Expected result: Test a renamed file, a new version, a moved folder, and an access-restricted document.
If it fails: Check the current file version and destination access before retrying transfer or metadata updates.
File or document metadata reporting workflow
Planning example. Stacksync support for the required connection and record operations needs a technical review.
Starting event: A change to the selected Proposed file or document metadata table in Azure Cosmos DB (choose its name) or Files record needs a defined result in the other system.
- Start with Azure Cosmos DB Proposed file or document metadata table in Azure Cosmos DB (choose its name) and Azure OpenAI Files. Use the record-matching and field-ownership rules from your mapping worksheet.
- Resolve folder/container and parent-record references before attaching metadata.
- Test a normal update and one failed or repeated update in the supported direction. Keep both record IDs with the test results.
Expected result: Test a renamed file, a new version, a moved folder, and an access-restricted document.
If it fails: Check the current file version and destination access before retrying transfer or metadata updates.
Model or deployment metadata reporting workflow
Planning example. Stacksync support for the required connection and record operations needs a technical review.
Starting event: A change to the selected Proposed model or deployment metadata table in Azure Cosmos DB (choose its name) or Models record needs a defined result in the other system.
- Start with Azure Cosmos DB Proposed model or deployment metadata table in Azure Cosmos DB (choose its name) and Azure OpenAI Models. Use the record-matching and field-ownership rules from your mapping worksheet.
- Resolve project, access policy, model version, and any evaluation requirements.
- Test a normal update and one failed or repeated update in the supported direction. Keep both record IDs with the test results.
Expected result: Test a retired model, a renamed deployment, and an unauthorized project reference.
If it fails: Confirm the current deployment version and access before retrying a dependent action.
Prepare approved Azure Cosmos DB data for AI workflows
This is an evaluation scenario; connector and operation support require confirmation.
Starting event: A selected Azure Cosmos DB record is approved for use in an AI workflow associated with Azure OpenAI.
- Define the minimum data from Items (JSON documents) required by the task and its access restrictions.
- Distinguish model/deployment metadata from actual inference or vector-index operations in Azure OpenAI. A sync connector does not establish those actions.
- Retain the source record ID and content version; define how corrections and deletions reach derived outputs.
Expected result: A corrected or withdrawn source record is reflected in the approved AI context; a record from another tenant is excluded.
If it fails: Rebuild derived content from the current authorized source. Do not retry an outdated action or assume model output can be written back without review.
Initial load and acceptance testing
Keep both record IDs with the expected and actual result. Reconcile the same filters and time window in each system.
Items (JSON documents) / Files
Test case
Test a renamed file, a new version, a moved folder, and an access-restricted document.
Expected result
The expected file or document metadata relationship is preserved with no duplicate action or unintended write.
Proposed file or document metadata table in Azure Cosmos DB (choose its name) / Files
Test case
Test a renamed file, a new version, a moved folder, and an access-restricted document.
Expected result
The expected file or document metadata relationship is preserved with no duplicate action or unintended write.
Proposed model or deployment metadata table in Azure Cosmos DB (choose its name) / Models
Test case
Test a retired model, a renamed deployment, and an unauthorized project reference.
Expected result
The expected model or deployment metadata relationship is preserved with no duplicate action or unintended write.
Direction and permissions
Test case
Bring an example source record and the intended destination operation to the compatibility review. Confirm the supported route before granting write access.
Expected result
Only an approved, supported direction and permitted fields are written.
Freshness and reconciliation
Test case
Measure source and destination times for the selected records under normal load and a burst. Reconcile IDs and values using the same filters and cutoff.
Expected result
The process meets its agreed freshness target and reconciliation has no unexplained differences.
Failed updates, retries and recovery
Start with the failed record and the destination error, then inspect the source value, field requirements, and access.
Rejected or repeated file or document metadata change
Investigate
Inspect Azure Cosmos DB Items (JSON documents) and Azure OpenAI Files, their IDs, and the destination error.
Next action
Check the current file version and destination access before retrying transfer or metadata updates.
Rejected or repeated file or document metadata change
Investigate
Inspect Azure Cosmos DB Proposed file or document metadata table in Azure Cosmos DB (choose its name) and Azure OpenAI Files, their IDs, and the destination error.
Next action
Check the current file version and destination access before retrying transfer or metadata updates.
Rejected or repeated model or deployment metadata change
Investigate
Inspect Azure Cosmos DB Proposed model or deployment metadata table in Azure Cosmos DB (choose its name) and Azure OpenAI Models, their IDs, and the destination error.
Next action
Confirm the current deployment version and access before retrying a dependent action.
A record type or update is unavailable
Investigate
Check the Azure Cosmos DB and Azure OpenAI connector guides, account permissions, and any operations marked On Request.
Next action
Ask the integration team to confirm a supported way to handle that record. Verify whether it needs connector configuration or a separate workflow step.
Source and destination disagree after a retry
Investigate
Compare current source values, destination validation, identity mappings, and any side effects already completed.
Next action
Stacksync issue retry reads the current source state. Decide the intended state before retrying or reverting; reconcile downstream effects separately.
Read the Stacksync issues dashboard guide for retry and revert behavior.
Change detection and update delivery
How updates move between Azure Cosmos DB and Azure OpenAI
See how each system detects changes and which updates the other system can receive. Each direction has its own permissions and record requirements.
Azure Cosmos DB Azure OpenAI Direction requires confirmation
Detect changesConfirm how Stacksync detects changes for this connector and the objects you need.
Apply updatesConfirm that Stacksync can create or update the records you need in Azure OpenAI.
Azure OpenAI Azure Cosmos DB Direction requires confirmation
Detect changesConfirm how Stacksync detects changes for this connector and the objects you need.
Apply updatesConfirm that Stacksync can create or update the records you need in Azure Cosmos DB.
Update timing and record limits
- Measure initial-load and ongoing-change latency separately. Source detection, selected objects, account limits, and destination validation determine the observed delay.
- Review write-back, deletion handling, update timing, and account limits with the integration team.
- Review write-back, deletion handling, update timing, and account limits with the integration team.
FAQ
Azure Cosmos DB and Azure OpenAI integration FAQ
-
Explore a Azure Cosmos DB and Azure OpenAI integration with a Stacksync engineer. Stacksync support for Azure Cosmos DB and Azure OpenAI is not established by the connector documentation reviewed for this page. Start with one record and the update your business needs to identify an implementation path.
-
Confirm two-way support with Stacksync for the records and fields you need in both systems. Access to a vendor API does not confirm that its Stacksync connector supports write-back.
-
Prepare both accounts, the selected object schemas, stable source and destination IDs, and the expected outcome. Identify the Azure Cosmos DB account, edition, environment, and business objects the integration must access. Identify the Azure OpenAI account, edition, environment, and business objects the integration must access. Use the pair worksheet to record ownership and acceptance criteria.
-
Measure initial-load and ongoing-change latency separately. Source detection, selected objects, account limits, and destination validation determine the observed delay.
-
No. Stacksync documents that pre-existing duplicates are not merged automatically when two-way sync begins. Review the initial dataset and matching plan before enabling it; an empty destination can simplify the first load.
-
Start with one business entity and a stable record ID. Map a small set of editable fields with compatible types, test required values and relationships, then expand after the pilot passes.
-
Check the destination error, field constraints, permissions, and current source value. The Stacksync issues dashboard supports retry and revert; retry reads current source values, so verify the intended record state before acting.
-
Use the current Stacksync pricing page and confirm the supported implementation with the team. Scope the required objects, record volume, update frequency, initial load, and support needs when comparing a managed connector with native or custom development.
-
Start by reviewing Items (JSON documents) in Azure Cosmos DB and Files in Azure OpenAI. Check how these records relate in your workflow, then confirm the actual fields and supported operations. Test record matching and one failed or repeated update before adding more records.
-
Choose a method around one example record and the update your business needs. Use Items (JSON documents) / Files to review record matching and confirm Stacksync support for the required operations. Compare ongoing sync, a custom workflow, and a scheduled export against that requirement.
Integration directory
Related integrations
Review documented support, sync direction, and setup requirements on each pair page. Search all 411 integrations listed for Azure Cosmos DB and Azure OpenAI.
Next step
Plan your integration with an engineer
Walk through your Azure Cosmos DB and Azure OpenAI records, field mappings, and requirements with an integration engineer.
